Odd problem with Sabertooth Z77

[BonzaiDuck]

First, I'm hoping that ASUS honors the "five year warranty" certificate, even for a second owner. But RMA may not be necessary. I'm awaiting their direction -- 48 hours from now.

I had a brand-new tested-good i7-2700K processor. I even had a modest $140 new motherboard for it, but I had been looking for better. Another enthusiast sold me his Sabertooth Z77 with an i5-3570K and 16GB of good RAM. The processor and RAM have been tested in the other board and are tip-top.

I put the SB-K processor in the board after resetting CLRTC. [C2032 battery is under thermal "armor" mobo duct -- will be a bitch to take apart so I can replace or unseat the battery.]

Board doesn't post, but shows the red RAM LED. I press the MemOK! button -- board goes through its configuration cycles. Finally, it posts with "memOK successful . . . F1 to enter setup." When I change the memory parameters to reflect stock DDR3-1600 9-9-9-24 2N and 1.50V, or XMP DDR3-1600 "Auto" timings and voltage, exit-save BIOS results in the same RAM LED light . . . press MemOK! . . . enter BIOS and the RAM is shown at 8-8-8-20, between 1066 and 1071 and 1.55V RAM voltage. If I leave the BIOS alone, the system will post without the MemOK frustration. But I can't set these RAM sticks (so far) to stock settings, or it just takes me through the same ol' MemOK cycle!

I'm wondering if:

1) Most recent BIOS version somehow overlooks compatibility with 2700K.
2) Some obscure setting I've missed.
3) The BIOS PLCC chip is borked.
4) Whether I should flash an older BIOS.

And -- finally -- this is now a second-hand board, no more than two years old, original owner volunteered the use had been "6 hours per day for a year." I got the 5-year-warranty certificate.

So I assume that if I can't fix this with observations 1 to 4, ASUS can. They'll either do it under the warranty coverage or they'll charge me something.

What I know: DDR3-1600 was NOT considered spec or stock and is technically "overclock" for i7-2700K. DDR3-1600 for 3570K IS considered stock. Both processors work another Z77 board, but I never thought to hook up the board with the original processor. I . . wanted . . . my . . . 2700K!!

Someone must have a similar experience or knowledge that will solve the problem. I even have a spare PLCC chip being shipped: I usually pick one up just for troubles like this, but it has most recent BIOS.

Maybe I should go ahead and flash the BIOS with an older version while I wait for ASUS to reply? Thoughts? Insights? Ideas?

 

[Meghan54]

Be aware that if Asus becomes aware that you're not the original purchaser, it can and most likely will deny warranty coverage.

ALL ASUS WARRANTY TERMS AND AGREEMENTS ARE NON-TRANSFERABLE AND ONLY APPLY TO THE ORIGINAL UNIT AND ORIGINAL PURCHASER.

http://www.service.asus.com/#!motherboard-warranty/c1c20

Section labeled Limitations of Liability, last para.

 

[VirtualLarry]

Does the RAM that you are attempting to use with your 2700K, have a 1333 SPD profile? If it doesn't, then that might explain what you're seeing.

 

[BonzaiDuck]

Does the RAM that you are attempting to use with your 2700K, have a 1333 SPD profile? If it doesn't, then that might explain what you're seeing.

I discovered the section of the Z77 BIOS that reports SPD -- something that had been added since Z68. The SPD info for the G.SKILL set that I swapped in seemed to be there. I just can't remember if there was a DDR3-1333 profile. I'll have to fire it up and look again . . . . then I'll come back here.

Also, I'm wondering if the "TPU" or whatever it us that drives the MemOK feature got "confused" when I popped in the new processor (perhaps before I thought to power-down, unplug and reset CLRTC.) I'm still wondering if the wafer-battery trick or replacing the PLCC chip will set it all straight. This is something we forget when we go goo-gah over a mobo-ducting feature like "TUF Armor" or whatever they call it. The mobo has to come out of the system, heatsink must be removed, 9x self-threading screws removed from the back, etc. etc. THEN you can fiddle with the wafer battery or replace the PLCC BIOS chip.

UPDATE: Yes, the SPD info appears in an "SPD information" menu. For the G.SKILL "GBRL" 2x4 kit in the system now, it only shows JEDEC @ 1600 with looser timings, and another SPD column @ 1600 with the spec timings.

But I'm trying to make sense of the implications. I KNOW that an i5-3570K's memory controller accepts DDR3-1600 as "stock" and not "overclocked" RAM; the 2700K only accepts 1333 as "stock," but NOBODY I ever knew had problems with putting the RAM in a Z68 board (or for that matter, the Z77-A board I have here as "spare") to get the MemOK feature to accept the full spec speed. The budget Z77-A board doesn't HAVE a MemOK feature.

Since I have to take the board out and remove the HSF just to RMA (in the event that I can do so), I can replace the wafer battery; I can even replace the PLCC chip (but with latest BIOS #2104) BEFORE I send it off. In fact, if there's a real problem for relying on ASUS to fix things, I'd have to do that as a last resort anyway.

Work, work, work. All work and no play makes Bonzai a dull duck. All work and no play makes Bonzai a dull duck. "Wendy! Wendy! Give me the bat! Give me the bat!" I'd like to get this Sabertooth up and running to prove its own "Shining," ha-hah! Before I go crazy . . . like Jack Torrance.

 

[BonzaiDuck]

Both memory kits -- the G.SKILL Ripjaws "GBRL" and the Corsair XMS3 -- have JEDEC SPD settings lower than DDR3-1600. So why aren't those listed I in the BIOS "SPD" tab? The tab only shows 1600 under both SPD and XMP#1 columns.

Could this have been an oversight in a "latest" BIOS update for a board more likely used for IB versus SB? Nothing was "done" to those RAM modules -- specifically the ones I already had before I got into these consequences -- to change that.

Also, as to simply "flashing" the BIOS with an older version, I tried that. The super-smart BIOS authors sought to reject earlier versions outright. So during the process of attempting to flash back to version 1304 (currently 2104) BIOS, a message box pops up saying the BIOS file to be outdated, with nothing more than an "OK" button to terminate the process. No option to go forward with it or override the new BIOS version with the old.

I need to acquire a version 1304 PLCC chip -- that's my tentative thought, anyway. Otherwise, it goes to ASUS, and if warranty coverage doesn't accommodate "second owner," I would have to pay. Something. . . I imagine. . . .

 

[BonzaiDuck]

I'm still waiting for ASUS tech-support to reply -- it takes 48 hours. I put in another ticket adding to my description of the first ticket: specifically that the JEDEC/SPD/XMP data shown in "SPD information" tab of the BIOS omits columns for DDR3-1333 or lower. I am intuitively accepting of the likelihood that this was an omission in the later BIOS revisions: the current version on the Sabertooth Z77 is 2104 -- confirmed by checking the main BIOS page.

Seller has communicated with my via e-mail to note that the EZ-Flash refusal to accept earlier BIOS revision can be over-ridden by the utililty "BUpdater" which I found in the SaberZ77's list of downloads.

First BIOS revision shown in the download list which mentions "compatibility with newer processors" seems to be version 1015. Before that, second BIOS revision is 0906. Now I'm wondering:

Will use of BUpdater result in catastrophe? [But I have the PLCC chip for 0906 on the way -- delivery likely late next week.]

Do I really need to go back that far?

IF no other insights to this problem, corrections to my hopeful assumptions etc., I will await response from ASUS tech-support.

[Maybe I should just replace the 2700K with the darn i5-3570K and see if that resolves the issue.]

By the way. Anybody remember how many times you can re-install one of these CPUs in a socket before pins and springs begin to wear out? Not eager to make changes just to "prove" something.

UPDATE ON COMMON SENSE:
This all started out with coveting a top-end SB-K chip. Then I wanted to see what it would do without reconfiguring my Z68 system after taking parting screen-shots of my config profiles and causing more trouble than I wanted. Now, I find this:

http://cpuboss.com/cpus/Intel-Core-i7-2700K-vs-Intel-Core-i5-3570K

Margin-of-error stuff!! And . . . pig-headed!!

What to do? What to do? Matched sets of PLCC chips and processors . . . I should've ventured into Devils Canyon. I am swimming in dated parts!! Need to set up my EBay store . . . :\

 

[mgttr]

DDR3-1600 was NOT considered spec or stock and is technically "overclock" for i7-2700K

I have a different board (Asus P8Z68 Deluxe/Gen 3) but have 4x4GB DDR3-1600 and an i7-2700K all working perfectly together for several years. All I did was select the XMP setting in BIOS for the RAM and that was the only tweak necessary. The CPU is not overclocked.

 

[BonzaiDuck]

I have a different board (Asus P8Z68 Deluxe/Gen 3) but have 4x4GB DDR3-1600 and an i7-2700K all working perfectly together for several years. All I did was select the XMP setting in BIOS for the RAM and that was the only tweak necessary. The CPU is not overclocked.

And it's reasonable to assume that the SB-K would be totally compatible in a Z77 motherboard. Reasonable -- because I have another cheaper board -- no probemo. No issue. And the cheap board was released late in the game: Asus always seems to come out with those "-A" models toward the end of the chipset life-cycle.

Truth is, after I got the 2700K processor originally slated for my sig-Z68 system, I became "hardware ambitious" thinking I could simply get another mobo and replace an LGA-775 system whose processor, RAM and board are promised to a friend and hardware junkie. It served its purpose. But it needs to be replaced anyway. It had "serious bidnis functionality," which requires it.

That was supposed to give me time to find the limits and sweet spots on the new Sandy, and manage a replacement without disrupting the sig configuration. Then I could just drop the 2600K into the box that had housed the LGA-775 Wolfdale system.

Anyway. ASUS tech got back to me in seeming agreement that the problem was the BIOS version which didn't adequately register the lower SPD spec required of the processor. But we've now run BIOS Flashback three times so far -- in what appears to be the tech-rep's strategy of binary search -- starting with the 7th of 13 BIOS revisions. The last flash installed the 2nd revision of the BIOS (going way, way back!). But while the SPD info for DDR3-1333 appears in the BIOS menu, the board and its MemOK! feature won't configure the RAM at more than 1066.

So there would be two paths of discovery here. Either the earlier BIOS was still INCOMPATIBLE with the RAM model (although the Corsair XMS3's had been around since Z68 days) and the problem was fixed in a different BIOS version I've yet to try; OR -- there is something wrong with either the MemOK! feature or the CMOS. There seems to be a spotty reporting of Advanced timing for the SPD column in the BIOS "SPD information" sub-menu -- once MemOK! completes.

In some ways, this entire saga seems absurd. Several "needs" like "replace server eventually, replace other household machines over the next year," as opposed to putting a more powerful processor in my Z68 board and calling it quits. Even if -- as I sought a used better-than-average mobo like the Sabertooth -- I also got an i5-3570K and 16GB XMS3 in the bargain, I want the Sabertooth to work. And not just work -- it has to work up to the expectations I'd have from a "new" retail box.

Do people really think that a motherboard can be "partially functional" and acceptable? Maybe -- if the problem is just a component like LAN port, an auxiliary storage controller or something that can be disabled and replaced with a PCI(or-E) card. But this is "digital" technology. Zero -- or One. It has to work, or it doesn't work.

 

[BonzaiDuck]

Looks like a bent pin in the lower quadrant, slightly left of center in the board. It is not the kind of bent pin that can be easily fixed without the proper equipment. I THINK (but I could check again) it is a light green VCCIO pin in the schematic.

I suppose I could keep diddling with it until the tech-support comes back via e-mail. Everyone talks about fixing bent pins, and I've fixed a few, but this one . . this one is a Doozy.

There is actually an even possibility that I did the damage myself, when I was first working with the board. I may have snagged something on the pin. the rest of them seem straight as a string.

 

[PhIlLy ChEeSe]

Looks like a bent pin in the lower quadrant, slightly left of center in the board. It is not the kind of bent pin that can be easily fixed without the proper equipment. I THINK (but I could check again) it is a light green VCCIO pin in the schematic.

I suppose I could keep diddling with it until the tech-support comes back via e-mail. Everyone talks about fixing bent pins, and I've fixed a few, but this one . . this one is a Doozy.

There is actually an even possibility that I did the damage myself, when I was first working with the board. I may have snagged something on the pin. the rest of them seem straight as a string.

Are you sure the board did not have a bent pin already, as it did come with the CPU? Any photos of the pins during the purchase?

 

[BonzaiDuck]

Are you sure the board did not have a bent pin already, as it did come with the CPU? Any photos of the pins during the purchase?

There was one pin only slightly out of kilter. For my shameful accident of chunking another pin with the corner of a processor PCB board when my hand slipped -- you know the rest. We'll see what they say about fixing it.

Just a thought to two cohorts of enthusiasts or users: younger folks who are in a hurry or impatient, and older folks feeling too fulla-beans and so confident that they let their guard slip. [Or . . . their shaky old hands slip . . ]

DON'T EVEN TOUCH THOSE PINS WITH YOUR FINGERS. If they're bent, use a fine-tipped mechanical pencil. Something like the point of a trapezoidal razor blade might be useful, but BE FREAKIN' CAREFUL!! If I bought a new board, I'd inspect the pins with a magnifier and strong light. And if there are seriously bent pins -- send the sucker back. You need a strong magnifier. If you have extra strong reading glasses, use them WITH the magnifier. If possible, obtain a jeweler's ocular. And if you still only see a blur of little gold surfaces -- don't proceed any further except to get an RMA ticket.

Don't DROP anything on the socket. Don't TOUCH the freakin' socket or pins. And always keep the socket covered with either a processor or the plastic socket-cover. A-MEN!!

 

[PhIlLy ChEeSe]

Best thing to use to see is a microscope, there not cheap either.....

 

[BonzaiDuck]

Best thing to use to see is a microscope, there not cheap either.....

No kidding. How you going to get the motherboard fitted to the slide base with the hole in it? I know! A secret "honey I shrunk the kids!" formula. But then, you need a higher-power-setting on the microscope!! :biggrin:

And you can't project light from below, either. I had a low-end medical-quality microscope when I was 13. What do they look like today -- excluding the electron variety?

 

[BonzaiDuck]

I started this thread, and I'm wondering if it could lead to another.

FIRST -- I CONFESS!! I was in DENIAL about it even as I went through the trouble of getting the Sabertooth to post. But denial is the mind's refusal to ACCEPT facts; it doesn't mean the facts didn't register in memory (human memory).

I CHUNKED a SOCKET-PIN, when I was installing the processor and my hand slipped. the corner of the CPU PCB hit that socket-pin with force. It jammed it down in a plastic crevice surrounding another pin, and it CREASED the spring part of the pin at the pin or nipple part of the metal.

This was the last effort I made at straightening a pin (spring . . . contact . . whatever). I ACTUALLY had the spring-with-nipple in place so all the nipples approximate lined up for contact with the gold processor contacts. I was trying to tease it into "perfect" alignment, with tension on the spring. The tool slipped; the spring flipped back; the nipple (or "pin") flew off at the crease.

MY FAULT!! Perfectly good motherboard; careless accident (1); careless perfectionism leading to accident (2).

So here's the deal. ASUS tech support has come back saying they can repair it. Whether they mean a single socket pin, or they'd replace the entire socket -- I can't say. It will cost something. Someone -- I won't say who -- guessed it to be $75. More, or less -- I am going to find out what it is. Since I probably got more than $600 in parts with the purchase of the board, I have a wider range of tolerance above that $75 possibility. Put it another way: with honesty and integrity, I could probably auction the processor and memory for two thirds of what I paid for the whole enchilada. That means -- on the basis of original price, age of components and proportionality -- I paid about $100 for a year-old ASUS SAbertooth Z77 board.

If it goes forward, I'll want to post a thread offering kudos for Asus tech-support. If they DO stuff like this, posting a testimonial will be good for all ASUS users.

 

[krnmastersgt]

The fee sounds like their socket replacement program they have in place for their top tier boards, something I came across looking at all their boards from working in retail.

 

[PhIlLy ChEeSe]

This sounds about right around $75 new socket, the board has a 5 yr warranty=WIN WIN. Done and over with.

Look forward to seeing some photos of the end result!

 

[BonzaiDuck]

This sounds about right around $75 new socket, the board has a 5 yr warranty=WIN WIN. Done and over with.

Look forward to seeing some photos of the end result!

Well, we're all S-O-L now. They came back and told me that it "required' entire replacement of the board. What they were asking was about $50 under the MSRP for the Sabertooth Z77 when it was a retail item.

Like I said elsewhere, though, I discovered today something I must have subliminally known about through word-of-mouth. I stumbled onto an offer at EBay for a certain board that mostly could only be found now through EBay and folks who "have to upgrade" and have only that justification to sell their boards.

There are companies out there that have two storefronts: One for their Buying customers and one for their "Selling" customers. They explain what they're doing in the second storefront, as they discuss taking charge of "IT assets" of other corporations. They test, they assess, and they move the inventory into two piles: One for resale on the other storefront, and one for their recycling business.

Spoke to a friend and retired electronics tech. Turns out he buys stuff from the company I found, and divvies it up among family. He swears by it. Notes that they had done similar business years ago buying surplus from the Gov, and were selling small refurbed gas engines for go-carts.

I think I have this troubled project covered. I lost the Sabertooth -- which was one third of a purchase that left me enough in good parts that I could simply break even on that alone. So I'm likely to have every 1155 processor covered with a board. And I can cut loose of both the socket and this nickel-dime hole I'm digging for reasons never anticipated before I had certain "accidents." If I'm short a board, there's no hurry about the surpluser's inventory. The options should be there for another year or two.

LET'S SAY that I got the Sabertooth Z77 for $100 considering the CPU and memory in the bargain. Each component sold new in a common price range. So I lost $100 for being a chucklehead and careless; I could get back the $100 pricing the other year-old parts at $50 below their current reseller price, or I could keep those parts. But the Sabertooth had a 10 phase-power design. I can get a board with 12-phase which I know well and with 22nm compatibility for $85.

I'm pretty much a chucklehead for not finding sources like this much earlier.

Fiddling with last-gen parts may be useful but troublesome. Finding those parts with some degree of assurance in reliability -- that's either a skill or an exercise in patience and common-sense.